Ryad Lotfi MAHTAL

- Took ownership of the entire data stack, leading data engineering initiatives end-to-end across architecture, orchestration, transformation, deployment, and analytics delivery. - Led the full migration of the Data Warehouse from PostgreSQL to ClickHouse, driving the project independently from technical assessment and design to implementation and production rollout - Evaluated costs, trade-offs, and performance implications of the new architecture, and delivered a production-ready platform that reduced dashboard loading time from 5 minutes to 5 seconds - Re-engineered several critical data workflows to improve reliability and maintainability, introducing unit and integration tests across transformation pipelines - Built and industrialized data models with dbt, orchestrated pipelines with Airflow, and supported deployment processes in Kubernetes-based environments.

- Designed and built multi-tenant REST APIs (FastAPI/Python) handling bulk data ingestion and search across MongoDB and PostgreSQL for major retail clients (Auchan, Carrefour, Maxeda, BUT) - Architected a data pipeline platform from end-to-end with Apache Airflow orchestrating 10+ ETL workflows processing diverse data sources (SFTP, APIs, CloudFlare, SAP) - Implemented event-driven architecture using RabbitMQ for real-time asynchronous data processing across multiple client environments - Integrated Apache Spark for large-scale data processing, optimizing execution times for high-volume SDO counter computations - Built a multi-client CI/CD system on GitLab enabling on-demand per-client deployments with automated credential generation and infrastructure provisioning - Developed a monitoring platform for Airflow, APIs, and RabbitMQ with Microsoft Teams alerting - Managed database infrastructure: PgBouncer connection pooling, PostgreSQL stored procedures, MongoDB aggregation framework and transactions optimization - Containerized all services with Docker and managed environments from development to production - Built and maintained an internal Python library (arislib) providing unified connectors for MongoDB, PostgreSQL, S3, Elasticsearch, and SFTP - Led international client integrations: Maxeda (Netherlands - Syndigo/SAP), Rona (Canada), and major French retailers

- Establishment of interface contracts to meet various business requirements; - Development of data flows on Azure Data Factory and Azure Databricks to migrate data from +10 subsidiaries worldwide (Luxembourg, Spain, Portugal, etc.). - Participation in the implementation of the Agile methodology; - Initiation of a git repository on AzureDevops and establishment of best practices for branch and code management; - Added CI/CD pipelines in Azure Devops to the project to test, build, and lint each pull request, - Data retrieval from multiple sources: Oracle DB, MySql, Web Scraping, etc. - Data storage on Azure Data Lake Gen 2 ; - Massive data processing using Pyspark and SparkSql on Azure Databricks; - Stream scheduling using VTOM ; - Code Review ; - Handling of production incidents

Comparative study between two Google Cloud services proposed for ETL: DataFusion & DataFlow - Design and provisioning of a BigQuery Data Warehouse for data analysis and storage - Implemented a Pipeline for extracting data from Google Cloud Storage, transforming the data into a custom format, and inserting it into a BigQuery table, using the Apache Beam SDK. - Implemented a custom Pipeline to support Slowly Changing Dimension using Apache Beam and Data Flow - Development of a BigQuery Connector for inserting, updating, deleting and retrieving data from BigQuery using the BigQuery SDK. - Documentation of the different GCP services, the different software architectures and the different analyses on Wiki.

- Engineered a large-scale traffic sign database for France, aggregating public datasets to simulate real-world driving conditions. - Developed a YOLO-based object detection model to localize traffic signs from image sequences with high accuracy. - Built a deep learning classification pipeline, combining CNNs and GANs for feature extraction, and MLP/SVM classifiers to predict traffic sign types from raw images. - Analyzed and optimized model complexity for real-time inference, applying network pruning techniques to reduce parameter count while maintaining performance.

Seminar conducted at Sorbonne University as part of the Master’s degree in Advanced Image Processing and Vision. The subject was Inpainting applied to oceanographic images using Generative Adversial Networks.

- Engineered a large-scale traffic sign database for France, aggregating public datasets to simulate real-world driving conditions. - Developed a YOLO-based object detection model to localize traffic signs from image sequences with high accuracy. - Built a deep learning classification pipeline, combining CNNs and GANs for feature extraction, and MLP/SVM classifiers to predict traffic sign types from raw images. - Analyzed and optimized model complexity for real-time inference, applying network pruning techniques to reduce parameter count while maintaining performance.

Un stage de découverte effectué au sein de la Multinationale Bonatti. Thèmes abordés : Instrumentation, Réseaux.

Chargé du bloc "Signal" au sein de CESI, l’École d’ingénieurs.

Un stage d'été effectué à l'Ecole Nationale Polytechnique en collaboration avec l'Université Paris-Sud. Le stage visait à concevoir une fusée à eau instrumentée. Mon rôle, en tant qu'électronicien, était d'étudier la possibilité d'instrumenter la fusée afin de récolter les différents paramètres du vol.

Thème : Capteurs et instrumentation.

Agenda Manager dans l'un des évènements les plus importants du club scientifique VIC de l'Ecole Nationale Polytechnique. The Engineer's Gala est un événement de réseautage qui vise à mettre en contacte les éléves fraichement gradués des différentes universités algérienne avec les représentants des entreprises nationales et multinationales.

Conception d'un système de localisation d'un robot autonome circulant dans une table de dimensions 3m*2m.

Thème : Maintenance des simulateurs de vols.

Membre de l'équipe Polybot. L’équipe qui a représenté l’École Nationale Polytechnique lors de la compétition internationale de robotique "Eurobot" en France. L'équipe Polybot a été vice championne du monde.